Ethanol Distribution, Dispensing, and Use: Analysis of a Portion of the Biomass-to-Biofuels Supply Chain Using System Dynamics

Transcription

1 Ethanol Distribution, Dispensing, and Use: Analysis of a Portion of the Biomass-to-Biofuels Supply Chain Using System Dynamics Laura J. Vimmerstedt 1 *, Brian Bush 1, Steve Peterson 2 1 National Renewable Energy Laboratory, Strategic Energy Analysis Center, Golden, Colorado, United States of America, 2 Peterson Group, West Lebanon, New Hampshire, United States of America Abstract The Energy Independence and Security Act of 2007 targets use of 36 billion gallons of biofuels per year by Achieving this may require substantial changes to current transportation fuel systems for distribution, dispensing, and use in vehicles. The U.S. Department of Energy and the National Renewable Energy Laboratory designed a system dynamics approach to help focus government action by determining what supply chain changes would have the greatest potential to accelerate biofuels deployment. The National Renewable Energy Laboratory developed the Biomass Scenario Model, a system dynamics model which represents the primary system effects and dependencies in the biomass-to-biofuels supply chain. The model provides a framework for developing scenarios and conducting biofuels policy analysis. This paper focuses on the downstream portion of the supply chain represented in the distribution logistics, dispensing station, and fuel utilization, and vehicle modules of the Biomass Scenario Model. This model initially focused on ethanol, but has since been expanded to include other biofuels. Some portions of this system are represented dynamically with major interactions and feedbacks, especially those related to a dispensing station owner s decision whether to offer ethanol fuel and a consumer s choice whether to purchase that fuel. Other portions of the system are modeled with little or no dynamics; the vehicle choices of consumers are represented as discrete scenarios. This paper explores conditions needed to sustain an ethanol fuel market and identifies implications of these findings for program and policy goals. A large, economically sustainable ethanol fuel market (or other biofuel market) requires low end-user fuel price relative to gasoline and sufficient producer payment, which are difficult to achieve simultaneously. Other requirements (different for ethanol vs. other biofuel markets) include the need for infrastructure for distribution and dispensing and widespread use of high ethanol blends in flexiblefuel vehicles. Citation: Vimmerstedt LJ, Bush B, Peterson S (2012) Ethanol Distribution, Dispensing, and Use: Analysis of a Portion of the Biomass-to-Biofuels Supply Chain Using System Dynamics. PLoS ONE 7(5): e doi: /journal.pone Editor: Guido Germano, Philipps-University Marburg, Germany Received June 27, 2011; Accepted March 13, 2012; Published May 14, 2012 Copyright: ß 2012 Vimmerstedt et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: All the funding was from United States Department of Energy, who had a role in the study design, data collection and analysis. Competing Interests: Steve Peterson is affiliated to the Peterson Group, West Lebanon. This does not alter the authors adherence to all the PLoS ONE policies on sharing data and materials. * laura.vimmerstedt@nrel.gov Introduction Transportation fuels from biomass (biofuels) are pursued to achieve multiple goals: reduced petroleum use and greenhouse gas emissions, fuel oxygenation, and agricultural market diversification [1,2]. This paper focuses on ethanol. Where real-world systems or modeled system representation is likely generalizable across all biofuels, we use the term biofuel to be broadly inclusive of all such fuels, including ethanol. In some cases we refer to ethanol or other biofuel to highlight that this analysis emphasizes ethanol but the statement in question applies broadly as well. We refer to ethanol for items that apply to that fuel only, and we refer to ethanol in presenting all results of the analysis because they are for ethanol specifically. The U.S. government has a long-standing goal of reduced dependence on imported petroleum that was initially prompted by the oil crisis and embargo of Federal incentives for the production and use of biofuels have been enacted as one method to achieve that goal. The Clean Air Act Amendments of 1990 (Public Law , 42 U.S.C. 7401) and associated fuel regulations established oxygenation requirements for much of the nation s gasoline supply, which oxygenated biofuels such as ethanol can help meet; these were subsequently supplemented with renewable fuel requirements [3,4]. More recently, the federal government has taken judicial, regulatory, and legislative steps to reduce greenhouse gas emissions. On April 2, 2007, the Supreme Court found in Massachusetts v. EPA, 549 U.S. 497 that greenhouse gases are air pollutants, resulting in regulatory action to limit greenhouse gas emissions from motor vehicles [5]. Climate change legislation has been introduced in Congress [6]. Low-carbon biofuels can help meet greenhouse gas mitigation goals. In acknowledgment of the potential contribution of biofuels to achieving these and other goals, recent policies at the state and federal levels in the United States have promoted the use of biofuels, with a long-term focus on ethanol and other biofuels. Such policies include: N a federal biofuels research, development, and deployment program N preferential tax treatment of biofuels production and sales PLoS ONE 1 May 2012 Volume 7 Issue 5 e35082

2 N subsidies for investments in fueling infrastructure N flexible-fuel vehicle (FFV) value for corporate average fuel economy compliance N renewable fuels standards [3,4,7]. Despite this history of policy intervention, biofuels use in the transportation fuel market remains small (10.7 billion gallons [8]) compared with the 36 billion gallons of biofuels by 2022 specified in the Energy Independence and Security Act of 2007 (EISA). Challenges to increased use include retail gasoline prices that are lower than the cost of delivering biofuel to the pump, logistics and infrastructure requirements for fuel distribution and dispensing, biomass-to-biofuel conversion cost and capital investment requirements, and logistical and market issues of biomass supply. Many of these challenges arise because the supply chain for biofuels differs substantially from the supply chain for petroleum-based transportation fuels [3,7]. Infrastructure-compatible biofuels (non-ethanol) would mitigate some, but not eliminate all, of these differences. To analyze such challenges, the U.S. Department of Energy (DOE) and the National Renewable Energy Laboratory developed a system dynamics modeling approach that represents the primary system effects and dependencies in the biomass-to-biofuels supply chain [9,10]. For purposes of this analysis, the biomass-to-biofuels supply chain is discussed at the overall industry level and with a focus on development and evolution of a supply chain for a developing industry. This is not to be confused with the day-to-day management of the existing supply chain of an individual firm working with other individual firms. In this context, the dynamics in question relate to the development of entire sectors of the industry on a year-to-year timescale (e.g. how long does it take production capacity to develop), not to supply chain management dynamics of individual firms that play out over weeks, days, or even hours (e.g. how long does it take this part to be shipped). This approach was designed to support biofuels policy analysis by determining what supply chain changes have the greatest potential to accelerate the deployment of biofuels (see Figure 1). In this paper, we address the downstream portion of the supply chain, including the distribution logistics, dispensing station, fuel utilization, and vehicle portions of the chain. The DOE-sponsored system dynamics model of the biofuels supply chain the Biomass Scenario Model represents major interactions and feedbacks related to a dispensing station owner s decision to offer ethanol fuel, with distribution options and vehicle choice represented as discrete scenarios. Figure 1. Overview of Biomass Scenario Model. The figure shows the overall purpose and content of the Biomass Scenario Model. doi: /journal.pone g001 This paper addresses the downstream section of the supply chain. The full supply chain is shown in Figure 2; the downstream end of the chain elaborates on the biofuels distribution and end-use components that are noted in the figure. Characteristics of the major components of the downstream system are summarized in Table 1. For each linked element in the downstream supply chain (i.e., entries in the first column), the infrastructure and market conditions necessary for an economically sustainable industry are listed, along with policies that can shape these infrastructure and market conditions. In the Methods section, we describe each of the supply chain links, infrastructure and market elements, and policy levers, as well as their representation in the model. In considering the downstream portion of the biofuels supply chain, three major analytic questions guide our exploration: 1. Can an economically sustainable domestic ethanol fuel distribution, dispensing, and end-use system develop, and under what policy and market conditions? 2. What combination of policies (including carbon policy) appears most likely to achieve policy goals? 3. How do these conclusions differ under different sets of assumptions about consumer and system behavior? We found that an economically sustainable domestic ethanol fuel distribution, dispensing, and end-use system could develop under conditions of strong policy intervention or significant change from current market conditions. Results suggest that, with sufficient policy intervention (or sufficient market change) across the supply chain, the policy goals could be achieved. However, policy intervention or market change relative to today s conditions would be necessary: competitive prices at the pump to attract consumers, financially adequate revenue streams to attract producers, and the infrastructure to bring the fuel to market are all necessary elements for a sustainable market, and do not exist at this time, nor are they envisioned in a business-as-usual situation. When assumptions about consumer behavior and system behavior differ, these conclusions generally persist, although substantially different assumptions affect the level of investment in incentives that would be required to develop and sustain an ethanol fuel system. The remaining sections of the paper will explore these questions, and possible answers, based on modeling results. Methods This Methods section describes each part of the biomass supply chain and briefly identifies what modules of the Biomass Scenario Model represent each part. The sections below offer more detail on how each part of the supply chain is represented. The Biomass Scenario Model is an analytic model that uses a system dynamics approach [11]. It is built in STELLA software [12]. The full model represents dynamic interactions in the biomass-to-biofuels supply chain, including the five interdependent systems shown in Figure 2: feedstock production, feedstock logistics, biofuels production, biofuels distribution, and biofuels end use [13]. This analysis uses a portion of the full Biomass Scenario Model, the downstream model. An analysis that uses the full model is underway and will explore the broader context for the downstream system addressed in this paper. The model is not intended to forecast the future; high-level models (such as the Biomass Scenario Model) are imprecise and best suited for analysis that focuses on relationships, interactions, and trends rather than on single-point PLoS ONE 2 May 2012 Volume 7 Issue 5 e35082

3 Figure 2. Supply chain for ethanol fuel with Biomass Scenario Model structure. The full supply chain for ethanol fuel is shown, with the downstream end labeled. The modules of the Biomass Scenario Model are briefly described. The downstream version of the Biomass Scenario Model that was used in this paper includes a simplified version of the Conversion Module and the Distribution Logistics, Dispensing Station, and Vehicle Modules. For simplicity, in the downstream version, the Vehicle Module is used to generate vehicle scenarios, rather than being fully coupled during each run. E10, E20, and E85 are light-duty vehicle fuels that are approximately 10%, 20%, and 85% ethanol by volume, respectively. doi: /journal.pone g002 predictions. Both the model development process and its results can help identify suspect assumptions and can thereby suggest agendas for further investigation. The downstream portion of the biofuels supply chain, and the corresponding downstream portion of the Biomass Scenario Model, encompasses a complex physical and market system that transfers biofuel from biomass-to-biofuel conversion facilities to its point of use as a transportation fuel. Table 2 shows selected assumptions that are used in this analysis in the downstream model. Biomass conversion to biofuel, which occurs at the upstream end of the chain, is beyond the scope of our analysis in this report. In order to depict the co-evolution of the downstream and conversion systems, we developed a simplified conversion module in the downstream model. Conversion is represented in greater detail in the full Biomass Scenario Model. Table 1. Infrastructure and Market Needs and Related Policies across the Downstream Supply Chain. Supply Chain Step Infrastructure Infrastructure Policies Market Conditions Market Policies Conversion Conversion capacity None Return on conversion investment Distribution Dispensing Tankage, terminal, truck, pipeline, rail capacity High-blend tankage and dispensing equipment Distribution and storage subsidy Repurposing subsidy; fixed capital investment subsidy End Use Flexible-fuel vehicles Manufacturer incentives; Car Allowance Rebate System ( Cash for Clunkers ) or other consumer incentives for vehicle purchase Return on distribution investment Return on investment in tankage and equipment Low relative end-use fuel price Point-of-production subsidy None None Point-of-use subsidy; incremental gasoline cost The table shows supply chain steps, infrastructure and infrastructure policies at each step, market conditions needed for ethanol market growth or stability, and market policies that apply at each step. doi: /journal.pone t001 PLoS ONE 3 May 2012 Volume 7 Issue 5 e35082

4 Table 2. Selected Assumptions. Item and Description Value Notes Maximum cellulosic feedstock production and corresponding ceiling for cellulosic feedstock price 500 million tons/yr; $105/ton The feedstock production is an input assumption based on EISA The ceiling price is an input consistent with that production level. Initial starch ethanol production capacity (Jan. 2006) gal/yr During simulation, production varies in response to price. Maximum starch ethanol production capacity 15 billion gal/yr Constraint on starch ethanol eligible for federal Renewable Fuel Standard under EISA Maximum cellulosic ethanol production capacity 44 billion gal/yr Based on feedstock production input assumption and an assumed 90 gal/ton conversion rate (500 million tons/yr690 gal/ton). Maximum ethanol imports gal/yr This ceiling is consistent with imports during the period when imports reached this level (but then collapsed). Initial ethanol point-of-production price (Jan. 2006) $1.50/gal During simulation, price varies to equilibrate supply and demand. Minimum point-of-production price for cellulosic $1.20/gal $2.40/gal For fully developed cellulosic industry. For initial-condition cellulosic industry. capacity addition Minimum point-of-production price for normal $1.00/gal $2.00/gal For fully developed cellulosic industry. For initial-condition cellulosic industry. cellulosic industry utilization Distribution and storage cost $0.25/gal $0.05/gal Input assumption for intra-region, no infrastructure. Input assumption for intra-region, with infrastructure. Cost of moving ethanol from one region to another $0.10/gal Input assumption for movement between any two regions. Transportation cost from distribution terminal to retail station $0.04/gal Input assumption for point-of-distribution to point-of-use delivery cost. Cost per station of repurposing existing tankage and equipment $20,000 Existing mid-grade equipment repurposed for high-blend fuel storage and dispensing. Cost per station of new tankage and equipment $60,000 Purchase and installation of new equipment for high-blend fuel storage and dispensing. The table shows some of the major assumptions that are used in the model. doi: /journal.pone t002 Increasing the availability of biofuels within the distribution system is a challenge. In the case of ethanol it must be kept separate from gasoline until the distribution terminal because of its chemical properties. At the distribution terminal, ethanol and gasoline are blended, and the blend is transported to a dispensing station (Figure 3). Truck or rail is generally used to transport ethanol from conversion facilities to distribution terminals, although at least one distributor, Kinder Morgan, uses a gasoline pipeline after careful testing to ensure adequate exclusion of water and oxygen [14]. In many cases, the current pipeline network is not appropriately located for ethanol or biofuel distribution. The likely regional distribution of domestic biofuel production is quite different from the regional population distribution and associated demand for transportation fuel, which exacerbates the challenge of developing a biofuel distribution system. In the Biomass Scenario Model, the distribution logistics module represents the distribution system (for more information, see the Distribution Logistics section). For ethanol or other biofuels marketed at retail, station owners must decide to offer the biofuel if it is to become more widely available at the pump. High-blend fuel is a blend of ethanol and gasoline with a high percentage of ethanol. An example of a highblend fuel is E85, which targets approximately 85% ethanol content by volume. High-blend fuel requires dedicated tanks, which represent a significant investment for station owners. Station owners also must consider risks of deciding to offer the fuel, such as risk of mishap or loss of sales during construction, liability risks associated with possible misuse of the fuel (i.e., if highblend ethanol were used in the wrong engines), and environmental contamination (i.e., if ethanol mobilized hazardous chemicals from leaking underground storage tanks). Whether or not the magnitude of these risks is significant, station owner perception of risk is a factor in their decisions. Dispensing stations operate in a very competitive market with thin profit margins and substantial shares of revenue from in-store sales. In this context, the choice to dispense biofuel does not lead to clear economic advantage at this time [15]. In the Biomass Scenario Model, the dispensing station module represents the dispensing system. We adopt an approach to consumer s choice based on rational choice theory and model price and availability of retail biofuels as determinants of consumer choice. While other factors play a role in choice, consumers are unlikely to seek out biofuels that are more expensive or require extra driving to obtain [15,16]. Complicating the understanding of price difference is the fact that ethanol has lower volumetric energy content than gasoline, and consumer awareness of this difference varies. The fuel utilization module of the Biomass Scenario Model represents consumer fuel choice. Major factors in the downstream system include availability of vehicles that can use fuel with high ethanol content; a consumer s choice to use the fuel; a dispensing station owner s choice to offer the fuel; and availability of the fuel within the distribution system. These factors would be of little or no significance for infrastructure-compatible biofuels that were blended with regular gasoline and not marketed separately at retail. Such fuels are being addressed in a new version of the Biomass Scenario Model that was not used in this analysis. Federal rules currently allow conventional gasoline vehicles to use fuel with up to 10% ethanol content [17]. EPA has partially granted a waiver that would allow use of gasoline that contains up to 15% ethanol in certain recent model years of vehicles. The current national level of ethanol use for transportation fuel is close to 10%. This national level is called the blend wall because this is the maximum amount that can PLoS ONE 4 May 2012 Volume 7 Issue 5 e35082

5 Figure 3. Ethanol distribution system. This conceptual view depicts the path of ethanol transport from biorefinery to dispensing station the ethanol distribution system. Storage capacity is included in the figure. Separation between ethanol and gasoline is generally maintained until the distribution terminal. The distribution logistics module of the Biomass Scenario Model represents the distribution system and represents expansion of that system due to supply push in which supply of ethanol coming out of the biorefineries that is in excess of distribution capacity prompts greater distribution capacity growth. Distribution within the production region (intra-region distribution) and from one production region to another (interregion distribution) is shown. doi: /journal.pone g003 readily be used under current rules. Higher levels of usage would require greater use of high-blend fuel in FFVs, conventional gasoline vehicles certified to a higher ethanol content fuel, biofuels other than ethanol that could be used in conventional gasoline vehicles at shares greater than 10%, or completion of federal rule revisions to allow up to 15% ethanol in gasoline. In the present version of the Biomass Scenario Model, a vehicle module that takes into account FFVs is used to estimate the potential for ethanol use in light-duty vehicles. Different vehicle mixes are explored as scenarios, but the vehicle module is not dynamically linked, and therefore is unresponsive to modeled changes in ethanol price. In the sections below, we define each element of the downstream supply chain, describe how it is represented in the model (including interactions and limitations), and describe policies that can be applied to each element and how those policies are represented in the model. Conversion Conversion is the transformation of biomass feedstock into biofuels. A full analysis of conversion using the Biomass Scenario Model is underway, but detailed analysis of conversion is not the focus of this study. For this study, we use a highly simplified representation that constrains the growth of conversion capacity while allowing for simple price feedbacks and pergallon subsidies at the point of production. This dynamic coupling to the downstream modules allows conversion capacity to grow in response to ethanol or other biofuel price and demand. The simplified conversion module represents the acquisition of production capacity for corn and cellulosic ethanol in each of the 10 agricultural regions employed by the Biomass Scenario Model. There are several important aspects to the logic of this simplified conversion module: N It represents both corn and a single cellulosic pathway for production of ethanol. (The more detailed conversion module captures multiple cellulosic pathways.) N A ceiling for cellulosic conversion capacity is set as a scenario. This level is consistent with feedstock availability, based on simulations from the feedstock supply module, which is not included in the downstream model. N Plant financials and investment decisions are captured implicitly using a Bass diffusion formulation [18]. N Feedback from price drives capacity acquisition and utilization. N The model uses a highly simplified representation of learning curve dynamics, in which the strength of price feedback on industry growth is a function of installed conversion capacity. Overall, then, the simplified conversion module represents growth trajectories for conversion capacity that are consistent with feedstock availability and responsive to price signals from downstream. Thus, this downstream analysis does not address questions, such as dynamics of changes in cost of production, that would require the detail of the full conversion module. Despite these limitations, the simplified conversion module permits representation of relationships between ethanol or other biofuel price and both the magnitude and the utilization of conversion capacity. For example, an increase in consumption can PLoS ONE 5 May 2012 Volume 7 Issue 5 e35082

6 drive the price of the biofuel upward. This price signal from the downstream module influences both the acquisition and the utilization of capacity, which in turn changes ethanol production, which feeds back to put a downward pressure on price. Representing this feedback loop an important market dynamic is a major purpose of the simplified conversion module. The pointof-use ethanol price in the downstream model is set to cover costs of delivering the fuel, plus an assumed markup ( cost-plus pricing). An alternative pricing representation would be to assume that if ethanol costs (including subsidies) are below gasoline price, that the ethanol price is set to match the gasoline price (marketbased pricing). A new version of the Biomass Scenario Model allows for the user to choose how to represent point-of-use ethanol price, with choices varying from the cost-plus price to the marketbased price. In addition to representing this feedback loop, the simplified conversion module allows for dynamic relationships among demand, interregional transport of biofuel, and relative levels of conversion capacity utilization across regions. If demand increases in a region with underutilized conversion capacity, conversion in that region will ramp up and transport of biofuel into that region will decline. Policies to advance conversion range from research and development to subsidies for conversion facilities. A full analysis of these policies is beyond the scope of this paper on the downstream system. In the downstream version of the model, representation of incentives for conversion is limited to a point-ofproduction subsidy. The point-of-production subsidy increases the price of ethanol or other biofuel that the conversion sector perceives, but does not directly affect fuel prices further downstream. This increases the incentive for investment in conversion capacity through the mechanisms described above by providing greater returns on conversion investment and by facilitating higher plant utilization rates at a given market price. Inclusion of this subsidy in the downstream model serves as a proxy for initiatives aimed at building industry capacity through per-gallon subsidies. In the full conversion model, the same price signal is used, but there it enters into an explicit calculation of the conversion facility investment decision that also includes a broader set of subsidies (feedstock, conversion facility output, and investment). Distribution Logistics In our analysis, distribution logistics is the set of organizational capabilities and infrastructure required to move biofuel from conversion facility to dispensing station. Generally, distribution logistics entails two steps: transportation from conversion facility to distribution terminal, and transportation from distribution terminal to dispensing station. The chemical characteristics of ethanol complicate distribution logistics because ethanol generally cannot be moved through pipelines used for petroleum products, stored in the same tanks as petroleum products, or blended with gasoline to make high-blend fuel until it is ready to be transported to the dispensing station. Mitigation of these limits is being explored. For purposes of this analysis, however, we assumed ethanol would require its own infrastructure for these steps and that the availability of high-blend fuels in the distribution system would therefore depend on development of ethanol distribution infrastructure not just on development of operational and regional conversion capacity [7]. In developing the distribution logistics module, the goal was to explore implications of distribution infrastructure, despite substantial uncertainty as to when, where, and with what component parts the infrastructure might develop. Rather than attempt to model detail about distribution infrastructure expansion, and then explore many scenarios to account for uncertainty, we developed a framework for capturing scenarios around the role of distribution infrastructure within the larger system. These are expressed as (1) different rates of infrastructure development based on evolution of conversion system and (2) cost implications of the evolving distribution infrastructure. The distribution logistics module categorizes distribution terminals as either possessing or not possessing the infrastructure required to handle ethanol in high volumes. The module provides a framework for estimating the costs of storage and transport of ethanol, both within and across regions. As ethanol infrastructure increases within a region, costs decrease for importing ethanol from other regions, for storing ethanol within the region, and for transporting ethanol within the region. The module is structured such that additional ethanol supplies in a given region will increase ethanol infrastructure penetration within that region first, and the additional supply will then encourage ethanol infrastructure additions in other regions (see Figure 3). This simplifying assumption is an imperfect representation of market behavior and could misrepresent regional flows when regional ethanol prices make other patterns of distribution more profitable. However, a more realistic representation was deemed overly complex for this version of the model. The distribution logistics module incorporates supply push pressure from the upstream side that could cause distribution terminals to add ethanol infrastructure capability. We do not represent the economics around the infrastructure investment decision. Instead, we simply assume that supply push pressure drives the acquisition of infrastructure over time. This rate of infrastructure acquisition is not instantaneous for a wide range of financial and physical reasons. In the model, the assumed rate of infrastructure acquisition can be changed to approximate the effect of more or less favorable financial conditions for investment, but any particular rate is not explicitly related to financial assumption. The point-of-distribution subsidy can be used to vary the constraints imposed by distribution infrastructure. The 10-region structure of the Biomass Scenario Model permits exploration of broad regional differences in ethanol distribution, but does not permit exploration of distribution logistics below the regional level. The model also represents ethanol or biofuel imports, which are being explored in other analyses. The primary interaction of the distribution logistics module with upstream modules is the supply push mechanism by which excess ethanol supplies motivate the acquisition of ethanol infrastructure at distribution terminals. The modules farther downstream do not directly feed price signals back to the distribution logistics module; instead, those price signals influence production, which in turn determines the amount of supply push. Because terminals that acquire ethanol infrastructure are assumed to retain it, no feedback is represented in the model that would cause terminals to stop offering ethanol once they had started. The distribution logistics module inputs to downstream modules by way of three variables: the availability of ethanol within the distribution system, a factor in the dispensing station owner s decision to offer highblend fuel, and the cost of distribution, which is a component of fuel price at point of use. Policies that encourage investment in distribution infrastructure could provide incentives that may be essential for its development. In the downstream model, these policies are represented by a pergallon subsidy to storage within the distribution system the distribution and storage subsidy and this lower cost is passed down the supply chain, ultimately reducing point-of-use price to consumers and possibly increasing demand. PLoS ONE 6 May 2012 Volume 7 Issue 5 e35082

7 Dispensing Stations Approximately 120,000 dispensing stations (as of 2007) provide ready access to vehicle fuel in most parts of the United States [19]. Dispensing stations obtain fuel by truck from the distribution network, store it in underground tanks, and pump it into consumers vehicles. Stations generally offer at least two grades of gasoline. In addition, a third, mid-grade gasoline, as well as diesel fuel, are widely available, and some dispensing stations offer other alternative fuels (for more information on alternative fuel availability, see stations/). Dispensing high-blend fuel requires significant investment at the dispensing station, including costs related to fuel storage tanks, pumps, management of legal and liability issues, and customer communications. Of these, the tankage investment is the most significant up-front issue, and stations can be categorized by the type of tankage investment needed. Some stations may have three different types of fuel tanks for gasoline one for each grade. Because pumps can blend low- and high-grade fuels to create midgrade fuels, the mid-grade tank at these stations can be repurposed for high-blend ethanol fuel. A station that does not have a midgrade tank that it could repurpose would have to install a new tank for high-blend fuel. Installing a new tank requires greater investment than repurposing one [15]. Costs to obtain dispensing infrastructure at a station could range from $2,500 at the lowest (for repurposing) to $200,000 at the highest (for installing new equipment in a high-cost location). Table 2 shows cost assumptions used in the model. Dispensing high-blend fuel offers station owners little certainty of increased profit: profit on fuel sales is low relative to profits on in-store inventory; increased sales of highblend fuel represent reduced sales of gasoline (overall, though not necessarily for each station); and the effect of high-blend fuel availability on in-store sales is uncertain. The dispensing station module represents, in considerable detail, the station owner s choice to invest in high-blend ethanol dispensing equipment. Because the model does not include vehicle choice, this decision is not part of a dynamic interaction of consumer vehicle choice, fuel availability, and distribution capacity. Given that FFV owners are often unaware of their vehicles flex-fueling capability and that FFV owners now refuel with high-blend only 4% of the time [8], this interaction is unlikely to be strong in the near term. The module first considers whether an individual station owner has access to ethanol in the distribution network. If so, the module represents the decision to consider installing ethanol-dispensing equipment. The module assumes that specific ownership categories (oil company owned, branded independent, unbranded independent, and hyper-mart stations) face different financial circumstances and market decisions. The module divides stations into those with repurposable mid-grade tanks and those without. A favorable net present value of the investment is a necessary condition for consideration. Net present value is calculated in a detailed pro forma that takes into account the expected net change in revenue and the investment cost associated with the investment in tankage and equipment. The calculation of change in revenue included low-blend sales, high-blend sales, in-store sales, and sales volume for each type of sale. The calculation of investment cost includes investment subsidies for fixed capital investment and repurposing subsidies, interest, taxes, and depreciation, and is discounted at different rates representing changes in required rate of return with change in depreciation or loan status. The point-ofuse subsidy improves the net present value by reducing the cost of high-blend. Other issues, such as competitive considerations, are also represented. The station owner is represented as having perfect knowledge of FFV ownership shares among current and prospective customers a simplifying assumption. The module explicitly calculates revenues from in-store sales, which account for a substantial share of station revenues and profits. In our base model assumptions, high-blend fuel, because of its lower energy content, results in more frequent fuel stops and slightly higher in-store sales; because this assumption is uncertain, we explore it in a sensitivity analysis (see Behavior and Market Sensitivity Analysis Results). Despite this detail, data to fully describe the station owner s decision process are limited, so the analysis incorporates sensitivities on several uncertain parameters. Other significant data limitations include inputs for tankage cost estimates. Cost is modeled as depending only on whether repurposing occurs or a new tank is installed, whereas actual installed costs of high-blend tankage are likely to depend significantly on other factors, such as whether station configuration makes tank repurposing or new purchase easy or difficult. The dispensing station module, like the distribution logistics module, interacts with other modules both by way of availability of high-blend fuel within the distribution system and via price. Greater availability of high-blend fuel dispensing capacity means that, in the fuel utilization module, more FFV owners have the option to choose high-blend. Retail pricing at the point-of-use is calculated as the cost of delivering ethanol or other biofuel to the dispensing station plus an assumed markup ( cost-plus pricing). An alternative pricing representation would be to assume that if biofuel costs (including subsidies) are below gasoline price, that the biofuel price is set to match the gasoline price (market-based pricing). A new version of the Biomass Scenario Model allows for the user to choose how to represent point-of-use biofuel price, with choices varying from the cost-plus price to the market-based price. The model seeks an equilibrium between production and consumption of biofuel in the entire supply chain without explicit representation of a retail-level market. If retail demand increases, lower inventory increases upstream price, ultimately feeding back to retail price. Because the station owners investment decision is represented in more detail than conversion and distribution investments, financial incentives for station owners can be represented in more detail. Policy levers within the dispensing station module include a subsidy for investment in new tanks (fixed capital investment subsidy), a tank-repurposing subsidy (repurposing subsidy), and a per-gallon subsidy to the price of fuel at point of use (point-of-use subsidy). These modeled incentives mirror the types of incentives that have been used to encourage industry growth. Fuel Utilization and Vehicles Combustion of ethanol or gasoline in light-duty vehicles is the fuel utilization considered here, and other vehicles and fuels are not considered. Low-blend ethanol fuel and high-blend ethanol fuel involve different decision makers. For low-blend fuel, utilization depends primarily upon the requirements for renewable fuels and limits on low-blend ethanol content (10% by volume is the current legal limit in the United States). The amount of ethanol used if all gasoline were 10% ethanol by volume is referred to as the blend wall. EPA estimates that the blend wall will be reached in 2014; the exact timing depends upon the rate of increase in ethanol use, the rate of decrease in gasoline use due to the increasing fuel economy of the light-duty vehicle fleet, and whether or not the legal limit for ethanol content is increased (e.g., to 15%). The blend wall does not apply to infrastructurecompatible biofuels. In contrast to pervasive use of low-blend, PLoS ONE 7 May 2012 Volume 7 Issue 5 e35082

8 utilization of high-blend fuel involves greater complexity, as it requires the availability of vehicles that can burn the fuel, the availability of fuel at the dispensing station, and consumers decisions to purchase the fuel. Current law allows high-blend E85 only in FFVs in the United States. Consumer fuel purchase decisions appear to be heavily influenced by, but not entirely dictated by, the price of high-blend compared with gasoline. The model estimates consumer choice of high-blend ethanol fuel. This estimate is made in the fuel utilization module of the downstream model. This module receives two key inputs from the dispensing station module: a vehicle scenario that determines the share of high-blend-capable vehicles and the availability of high-blend fuel at the dispensing station. It accounts for the regulatory requirements encouraging use of low-blend fuel and estimates consumer choice of high-blend. For FFV owners with access to high-blend dispensing, the fuel utilization module models consumer choice between high-blend and gasoline. Although additional refinements might be more realistic, the module now represents consumer decisions based on the price per unit energy of fuel not volume of fuel and it assumes consumers have complete knowledge of availability of high-blend fuel at dispensing stations. Consumers are simplified into two categories: If gasoline and high-blend fuel were identically priced on an energy basis and available, regular users would choose high-blend fuel 80% of the time, while occasional users would choose gasoline 80% of the time. The Biomass Scenario Model does have a vehicle module, although it is not dynamically coupled to the downstream model for this study. The vehicle module does not incorporate consumer vehicle choice but instead provides an accounting framework in which to examine potential policy influences on vehicle stocks and associated maximum potential ethanol utilization. The vehicle module can generate policy scenarios that can be used in the downstream model. The pattern over time of maximum ethanol consumption potential under each scenario is an input to the integrated downstream model. Accordingly, the model in its current form does not represent vehicle choice and related interactions and is limited in its representation of consumer choice of fuel. It does not represent dynamic interactions of fuel price and fuel availability with vehicle choice, nor does it estimate effects of education and outreach efforts on vehicle choice. Dynamic interactions could become a more important consideration if FFVs were refueled with highblend a greater share of the time and if consumer choice played a greater role in FFV purchase. The model also does not examine consumer fuel choice among atypical consumers (e.g., those who would pay a substantial premium or drive far out of their way to obtain high-blend fuel or those who would decline high-blend fuel under any circumstance). Within the current model structure, because data on consumer fuel choice are limited, we conducted sensitivity analysis on the assumptions about choices of regular users and occasional users (see Behavior and Market Sensitivity Analysis Results). Policies related to fuel utilization include any policy that lowers the price of ethanol at the pump relative to the price of gasoline and to upstream ethanol prices. Such policies could include carbon taxes or cap policies, gasoline taxes or floor prices, ethanol subsidies or ceiling prices, or other incentives. In the downstream model, these can be represented as gasoline taxes or high-blend fuel subsidies. The increase in cost of high-blend fuel if a carbon tax or a carbon cap policy were implemented would likely be smaller than the increase in the cost of gasoline, although both fuels emit carbon. The downstream model also includes a point-ofuse subsidy, which reduces ethanol price at the pump. Vehicle policies such as the Car Allowance Rebate System (Cash for Clunkers), vehicle purchase incentives, FFV standards, and efficiency standards can be represented as different maximum potential ethanol consumption scenarios in the downstream model, based on outputs from the vehicle module. Structural Summary of Downstream System The downstream system exhibits the challenge of maintaining a stable, competitive ethanol fuel market. Gasoline is the dominant fuel in the light-duty vehicle market, and the cost of driving a gasoline-fueled vehicle sets a ceiling above which alternative vehicle-fuel systems are unlikely to compete for substantial market share. On the supply side, ethanol at point of production must sell at a price sufficient to compensate upstream market actors agribusiness, feedstock logistics operators, and ethanol conversion facilities for their investments and labor. Overall, the behavior of the downstream system reflects the complex feedback between supply and demand, which connect producers and consumers at each stage of the supply chain. The downstream system also demonstrates the need for coordinated growth across the entire downstream supply chain. Insufficient capacity at any link will inhibit growth of the market overall, as indicated in the results in the Results and Discussion. Different parts of the system respond to signals for investment with different lag times, complicating the necessary coordination in growth across the system. Results and Discussion We used the Biomass Scenario Model to explore how incentives for market actors across the supply chain shape the potential for a sustainable ethanol market in the United States. In this section, we present the results of model runs of the downstream system that were designed to explore these issues. A supporting information file offers a data supplement (Data S1) that includes results of all of the model runs that are presented here, as well as additional model runs. The results displayed in Figure 4 show simulated actual ethanol consumption from several modeled cases: a No Policy case without incentives (not even current incentives), a Higher Market and Infrastructure case with higher levels of incentives, both on a per-gallon basis and for infrastructure investments, and a Lower Market and Infrastructure case with lower levels for these incentives. Incentive levels for these cases are shown in Table 3. Note that none of these cases is intended to represent continuation of current policy. We recognize that the levels of policy intervention explored in some of these cases are significant; this paper does not advocate adoption of these particular policies, nor do we assert that the results presented here are accurate projections of the cost and effect of these policies. Instead, we intend to explore the behavior of the system as modeled, seeking insights on system behavior, not to predict specific numerical results. Also shown in the figure, for comparison, is an external dataset that does not represent model results: The EISA Legislated Total Renewable Fuel Requirement represents a growth trajectory that would rise to the 36 billion gallons per year of biofuels utilization by 2022 that is targeted in the legislation. The EPA Administrator has discretion to reduce the required level if the targets proposed by Congress cannot be met. Congress has not set production targets for years beyond Ethanol fuel in light-duty vehicle markets is only one of the pathways to meet this total requirement, but the total is offered for reference here because the legislation does not establish a total level of ethanol or a total amount of lightduty vehicle biofuel. PLoS ONE 8 May 2012 Volume 7 Issue 5 e35082

9 Figure 4. Scenarios for actual and potential ethanol consumption. The figure shows ethanol consumption results for all years from three modeled scenarios: a No Policy case (i.e., no incentives), a Higher Market and Infrastructure Incentive case, and a Lower Market and Infrastructure Incentive case. For comparison, results for maximum potential consumption ethanol consumption with a particular vehicle scenario are also shown (Maximum Potential Consumption under Vehicle Base Case); the vehicle scenario is based on the Energy Information Administration s Annual Energy Outlook [20]. An external dataset, not representing model results, is also shown for comparison: the EISA Legislated Total Renewable Fuel Requirement, which shows the goals of Energy Independence and Security Act of 2007 for renewable fuel (not necessarily ethanol). The figure illustrates that model results suggest that reaching EISA targets would require considerable incentives. doi: /journal.pone g004 The results shown here indicate that, under the modeled conditions, reaching ethanol consumption, even well below EISAlegislated targets for renewable biofuels, appears to require policy intervention with substantial overall financial implications. The Maximum Consumption under Vehicle Base Case Scenario uses the maximum potential ethanol consumption (i.e., ethanol usage if all the flex-fuel vehicles use high-blend fuel exclusively)implied by a vehicle scenario in this case, the number and type of vehicles in Energy Information Administration s Annual Energy Outlook reference scenario [20]. In Behavior and Market Sensitivity Analysis Results, we explore how vehicle policy scenarios would change this potential consumption of ethanol. Table 3. Summary of Incentives. Policy Category Incentive Name Range of Values Considered Definition and Notes Market Incentives Point-of-production subsidy $0/gal $1/gal For cellulosic ethanol producers only. Point-of-use subsidy $0/gal $1/gal Lowers ethanol price at point of use. Fiscal effect increases with increasing ethanol use. Incremental gasoline cost $0/gal $2/gal Represents any policy or market condition that differentially increases gasoline price at point of use (e.g., carbon policy, high oil price with little ethanol price response). As a tax, the fiscal effect decreases with increasing ethanol use. Infrastructure Incentives Distribution and storage subsidy $0/gal $1/gal Subsidy for tankage and equipment at distribution terminals and for distribution. Repurposing subsidy 0% 75% Percentage of total cost that is subsidized. Subsidy for repurposing of existing dispensing station infrastructure. Fixed capital investment subsidy 0% 75% Percentage of total cost that is subsidized. Subsidy for installation of new dispensing station infrastructure. The table summarizes the market and infrastructure incentives, the range of values for the incentives that are considered in the analysis, and further details about each incentive. doi: /journal.pone t003 PLoS ONE 9 May 2012 Volume 7 Issue 5 e35082